a conceptual issue: Is there a difference between estimating heritability by using GCTA and estimating variance explained by common SNPs?
I am asking because I am interested to know if I should use the filter of close relatedness (>0.025) or not? If I don't exclude those relatives (closer than 3rd cousins), will I have a problem estimating the narrow sense heritability in my sample? Or is this filter should be used only in the case I want to estimate proportion of variance explained by common SNPs?
Yes, there is a difference. Please see FAQ #5. www.complextraitgenomics.com/software/gcta/FAQ.html In brief, GCTA estimate using family data including close relatives is similar to (not exact the same) the heritability estimated from traditional family and twin studies, whereas the estimate using unrelated samples is an estimate of variance explained by SNPs rather than the trait heritability.